Kate Hughes, Jacky Croke, Rebecca Bartley, Chris Thompson, Ashneel Sharma
PII: S0169-555X(15)30091-X
DOI: doi:10.1016/j.geomorph.2015.07.024 Reference: GEOMOR 5317
To appear in: Geomorphology
Received date: 9 December 2014 Revised date: 13 July 2015 Accepted date: 15 July 2015
Please cite this article as: Hughes, Kate, Croke, Jacky, Bartley, Rebecca, Thompson, Chris, Sharma, Ashneel, Alluvial terrace preservation in the Wet Tropics, northeast Queensland, Australia, Geomorphology (2015), doi: 10.1016/j.geomorph.2015.07.024
This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
ACCEPTED MANUSCRIPT
Alluvial terrace preservation in the Wet Tropics, northeast Queensland, Australia
Kate Hughes1*, Jacky Croke1,2, Rebecca Bartley3, Chris Thompson1, Ashneel Sharma2
1
School of Geography, Planning and Environmental Management, The University of Queensland, Brisbane 4072, Australia
2
Department of Science, Information Technology, and Innovation, Brisbane 4068, Australia
3
CSIRO, Brisbane 4068, Australia
ACCEPTED MANUSCRIPT
AbstractAlluvial terraces provide a record of aggradation and incision and are studied to understand river response to changes in climate, tectonic activity, sea level, and factors internal to the river system. Terraces form in all climatic regions and in a range of geomorphic settings; however, relatively few studies have been undertaken in tectonically stable settings in the tropics. The preservation of alluvial terraces in a valley is driven by lateral channel adjustments, vertical incision, aggradation, and channel stability, processes that can be further understood through examining catchment force-resistance frameworks. This study maps and classifies terraces using soil type, surface elevation, sedimentology, and optically stimulated luminescence dating across five tropical catchments in northeast Queensland, Australia. This allowed for the identification of two terraces across the study catchments (T1, T2). The T1 terrace was abandoned ~13.9 ka with its subsequent removal occurring until ~7.4 ka. Abandonment of the T2 terrace occurred ~4.9 ka with removal occurring until ~1.2 ka. Differences in the spatial preservation of these terraces were described using an index of terrace preservation (TPI). Assessments of terrace remnant configuration highlighted three main types of terraces: paired, unpaired, and disconnected, indicating the importance of different processes driving preservation. Regional-scale variability in TPI was not strongly correlated with catchment-scale surrogate variables for drivers of terrace erosion and resistance. However, catchment-specific relationships between TPI and erosion-resistance variables were evident and are used here to explain the dominant processes driving preservation in these tropical settings. This study provides an important insight into terrace preservation in the tectonically stable, humid tropics and provides a framework for future research linking the timing of fluvial response to palaeoclimate change.
ACCEPTED MANUSCRIPT
Keywords: narrow valleys; partly confined; stream power; tropical rivers;
ACCEPTED MANUSCRIPT
1. Introduction
Alluvial terraces are topographic landforms that result from floodplain abandonment when a river incises, therefore providing records of river aggradation and incision (Leopold et al., 1964). Terraces have formed in all global regions and are commonly studied to understand how rivers have responded to the combined effects of changes in external controls including climate, tectonics, and sea level (Merritts et al., 1994; Macklin et al., 2002; Bridgland and Westaway, 2008) and in internal controls, such as catchment morphology (Coulthard et al., 2005) and reach-specific conditions (Houben, 2003).
Terraces are generally identified in the landscape using a range of techniques including soil properties (see reviews in Birkeland, 1990; Huggett, 1998), topographic data (Pazzaglia and Gardner, 1993; Jones et al., 2007), aerial and satellite imagery (Litchfield and Berryman, 2005), with correlations further aided by sedimentology and dating (Stokes et al., 2012). The use of soil characteristics to classify terraces has made an important contribution to understanding the broad temporal patterns of terrace abandonment (Bull, 1990). Following channel incision, time-dependant weathering processes lead to the development of pedogenic constituents such as clay and iron minerals, and thus soil characteristics are useful for broadly differentiating the ages of terraces (Walker, 1962; Harden and Taylor, 1983; Tsai et al., 2007). In this way, the degree of soil development can help overcome issues associated with terraces in areas of subdued topographic relief where differences in surface elevations can be unclear (Warner, 1972; Cohen and Nanson, 2008).
The identification of terraces has predominantly been for the purposes of studying valley aggradation phases (i.e., terrace production), with much less attention directed toward the erosional processes that influence terrace preservation (Lewin and
ACCEPTED MANUSCRIPT
Macklin, 2003). Terrace preservation, which is represented by the remaining terrace as a proportion of the valley floor, reflects changes in the processes of channel incision, lateral removal, channel aggradation, and channel planform stability (Colman, 1983; Lewin and Macklin, 2003). These factors are, in turn, influenced by catchment characteristics such as contributing area (e.g., Lewin and Gibbard, 2010), valley width (e.g., Richardson et al., 2013), stream power (e.g., Fryirs and Brierley, 2010), and bedrock resistance (e.g., Reneau, 2000).
The factors that influence terrace preservation can be represented using the concept of force and resistance (Bull, 1979). Using this concept, fluvial response (i.e., terrace removal) can be considered to reflect the balance between driving factors that promote their erosion (e.g., stream power) or resistance (e.g., boundary material resistance). While Bull (1979) suggested the use of the specific variables of stream power to represent force and critical power to represent the resistance, a broader, catchment-scale focus may encompass other factors such as catchment area, drainage density, slope, lithology, stream power, valley width, or alluvial material properties that are relevant to processes operating over timescales of terrace evolution.
The main aim of this paper is a spatial classification of terraces across five catchments in the Wet Tropics region of northeast Australia for the purposes of examining terrace preservation and catchment factors associated with their removal. In this paper we undertake desktop mapping of terraces based on terrain, elevation, and soil type and validation using ground truthing and independent chronostratographic data. We assess catchment terrace preservation using metrics for spatial preservation and remnant configuration. We also investigate the relationship between terrace preservation and a range of catchment variables selected to represent drivers of terrace erosion and resistance (i.e., the force-resistance framework).
ACCEPTED MANUSCRIPT
Collectively, these analyses are used to present a conceptual model to explain the dominant terrace types and the processes driving their preservation in the Wet Tropics. This study provides a framework for additional research linking the timing of fluvial response to palaeoclimatic change in the Wet Tropics.
2. Regional setting
The Wet Tropics biogeographic region (~22, 000 km2) on the east coast of far north Queensland is situated between 15oS and 19oS and comprises 11 catchments that drain the Great Dividing Range to the Coral Sea (Fig. 1). It has a wet tropical climate and is among the highest rainfall regions in Australia, with average annual rainfall varying from 1700 to 3000 mm on the coastal plain to 5000 to 7500 mm on the higher ranges and peaks (McJannet et al., 2007). There are steep east to west declining rainfall gradients as well as latitudinal variability largely owing to the orientation and arrangement of the coastal ranges (Bonell and Gilrnour, 1980). The majority of the Wet Tropics’ alluvial valleys are located within a region that receives about 2000 mm in mean annual rainfall. Rainfall is concentrated between December and March in association with the southward migration of the Austral monsoon trough. The largest sources of rainfall are tropical depressions and cyclones within the monsoon trough (Bonell, 1988), which typically produce rainfall of high intensity and long duration with daily rates of 8 to 27 mm h-1 (Bonell and Gilrnour, 1980).
ACCEPTED MANUSCRIPT
Fig. 1. Location map of (A) Wet Tropics region in northeast Australia and selected study catchments of (B) the Daintree River and (C) the North Johnstone River, South Johnstone River, Liverpool Creek, and Tully River. The inset boxes highlight the focus study reaches in each catchment.
ACCEPTED MANUSCRIPT
There are three main geologic units in the Wet Tropics: (i) Silurian-Devonian metamorphic sediments of the Hodgkinson's Formation; (ii) Carboniferous–Permian granites that intruded into the older metamorphics and form the coastal ranges and peaks (1000-1600 m Australian Height Datum - AHD); and (iii) Miocene and early Quaternary basalts formed in the vicinity of Atherton and flowed down into the present catchments of the North and South Johnstone Rivers (Willmott and Stephenson, 1989). The region is considered to be relatively tectonically stable (Geoscience Australia, 2015). Extensive weathering of bedrock has led to the formation of fine-grained regolith mantles, ranging from 4 to 6 m thick, on slopes <20%, on all lithologies (Thomas, 2004).
2.1. Selected study catchments
This study focuses on terraces in the Daintree, North Johnstone, South Johnstone, Tully, and Liverpool Creek catchments (Fig. 1). These catchments were selected to represent a range of variables such as catchment area, geology, channel gradient, and valley width and also span the pronounced north-south rainfall gradient in the region (Table 1). All five catchments are characterised by a progression from bedrock-dominated gorges to narrow valleys and partly confined rivers with discontinuous floodplains and terraces, to a coastal plain with largely unconfined channels and fluvial and estuarine deposits. Each catchment represents contributions of the three major geologies present throughout the region (Table 1). The Daintree River rises in the granite tablelands 1348 m AHD, but the majority of the river (85 km) flows as a confined channel through narrow valleys underlain by Silurian-Devonian metamorphic rocks. The North and South Johnstone Rivers traverse basalts on the Atherton Tableland (1150-1350 m AHD) and flow through narrow gorges (20
ACCEPTED MANUSCRIPT
km in length) to exit into narrow valleys that extend for 25 km downstream. Liverpool Creek rises at the escarpment edge (1145 m AHD) on Hodgkinson’s metamorphic rocks, while the headwaters of the Tully River traverse Carboniferous-Permian granite ranges (1060 m AHD) and acid volcanics. This study focuses specifically on alluvial sequences preserved within 15-20 km long reaches in narrow valleys <3 km wide, located between 10 and 40 km upstream of the river mouth, herein termed study reaches (Fig. 1).
3. Methods
In order to contrast terrace preservation across the five catchments, terraces were identified (through mapping and classification) during an initial desktop analysis phase using data on terrain, longitudinal profiles, and preexisting soils mapping. This data was subsequently validated using sedimentology and optically stimulated luminescence (OSL) dating. Collectively these data were used to inform a regional-scale correlation of terraces across the five catchments. Correlated terraces were allocated a sequence code (e.g., T1, T2) with T1 denoting the highest terrace observed across the region. Preservation and configuration of terrace sequences in each catchment were assessed in order to (i) examine the relationships between terrace preservation and surrogate catchment variables for terrace erosion or resistance and (ii) determine processes of terrace removal.
3.1. Terrace identification and correlation 3.1.1. Spatial extent
Terraces were identified based on the presence of a tread (i.e., a flat, elevated surface) that is separated by a distinct break in slope of >20o from lower terraces or
ACCEPTED MANUSCRIPT
the floodplain, which is inundated by annual monsoon floods. Terrace remnant perimeters were digitised from Light Detection and Ranging (LiDAR) derived 1-m digital elevation model (DEM) and its derivative slope and hillshade grids in ArcMap 10.1 (ESRI, 2012). The spatial extents of the mapped terraces were validated in the field.
3.1.2. Longitudinal profiles
To classify the downvalley relative elevations of terrace and floodplain surfaces and channel profiles, the mean surface elevation (m AHD) for each terrace remnant tread and adjacent floodplain and channel was extracted from the LiDAR DEM and longitudinal profiles generated for each study reach.
3.1.3. Soil type
Soil characteristics can inform the broad temporal differences in terrace abandonment. In this study, preexisting soil type mapping was used to assess the longitudinal profiles of different terrace surfaces by association with specific soil types. Soils mapping was undertaken across the Wet Tropics catchments between 1986 and 1996 using standardised field soil survey techniques (scale 1:50, 000; available at https://publications.qld.gov.au/). Soils were analysed to a depth of 2 m at an approximate site density of 2 to 4/km2 (CSIRO, 2008). Soil types were classified based on (i) parent material, (ii) landscape position, and (iii) the soil physical and chemical characteristics according to standard methods (Isbell, 1996; NCST, 2009). The physical and chemical properties are described in published reports (Murtha, 1986, 1989; Cannon et al., 1992; Murtha et al., 1996). Digital soil data layers were compiled in ArcMap 10.1 and overlain on terrace mapping to identify the soil types
ACCEPTED MANUSCRIPT
associated with each terrace remnant. The soils data were used to highlight the occurrence of multiple soil types within single longitudinal profiles for subsequent validation.
3.1.4. Sedimentology
To verify the association between soil types and the underlying stratigraphy and sedimentology, cores were extracted using a GeoProbe drill rig from terrace fronts at several representative locations within each of the study reaches. Field texturing and pedological characteristics such as Munsell colour, pedologic structure, and mottling were recorded down profile. Once returned to the laboratory, cores were logged for major sedimentary facies after Miall (1985); and where possible, they were grouped into major stratigraphic units, and the major element boundaries were identified (sensu Brierley, 1996).
3.1.5. Dating
A broad terrace chronology is provided using samples extracted from near-surface and basal facies from terraces and floodplains with the aim to (i) verify soil types as broad indicators of terrace age and (ii) constrain the timing of terrace abandonment. Dating of terrace or floodplain basal sediments, or the material deposited immediately above, is considered representative of the timing for the onset of infilling. Where near-surface sediments are dated, ages are considered to be minimum terrace/floodplain ages. Minimum near-surface ages can only represent the earliest age of abandonment because of the possible removal of younger sediments through erosion of the terrace surface. However, when combined with basal ages from
ACCEPTED MANUSCRIPT
an adjacent lower terrace or floodplain, the age of abandonment and period of removal can be constrained.
Material above 1.0 m was not sampled because of potential complications of bioturbation (Duller, 2008). Material suitable for OSL dating was extracted from cores under subdued red-light conditions (>590 nm) and processed to isolate pure extracts of 180-212 μm light-safe quartz grains following standard procedures (Aitken, 1998). Single-grain/small aliquot (two to three grains) equivalent dose (De)
values were determined using the modified single aliquot-regenerative dose (SAR) protocol of Olley et al. (2004) and Risø instrumentation described therein, in combination with the acceptance/rejection criteria of Pietsch (2009). We followed an established approach for age modelling (Galbraith and Laslett, 1993; Galbraith et al., 1999; Roberts et al., 2000), which involves the use of (i) the central age model to define the degree of overdispersion in single grain De populations and (ii) the
minimum age model to identify the De component pertaining to fully bleached grains.
Lithogenic radionuclide concentrations were analysed from sediments adjacent to OSL samples using high-resolution gamma spectrometry for 238U, 226Ra, 210Pb, 232Th, and 40K concentrations (Murray et al., 1987). Dose rates were calculated using the conversion factors of Stokes et al. (2003) and β-attenuation factors from Mejdahl (1979). Cosmic dose rates were calculated from Prescott and Hutton (1994) using an alpha-efficiency ‘α ’ value of 0.04 ± 0.02 (Bowler et al., 2003). Water contents were measured directly from each sample and have been assumed to be representative of saturation levels over the full period of burial, assigned errors of ± 5% to account for uncertainty.
ACCEPTED MANUSCRIPT
3.2. Terrace preservation assessmentsTo determine the extent and processes of terrace removal, terrace preservation was assessed in terms of the remaining spatial area using an index of terrace preservation, terrace surface dissection, and cross- and downvalley configurations of terrace remnants.
3.2.1. Terrace preservation index
The terrace preservation index (TPI) describes the preserved area of a terrace sequence as a ratio of the assumed original terrace area (Colman, 1983) as
TPIx = A/ M (1)
where x represents the terrace sequence code (e.g., T1, T2), A is the current preserved terrace area of the terrace sequence, and M is the assumed area prior to downcutting and abandonment. In the case of the T1 terrace, the assumed original terrace area was set equal to valley floor area (Leopold et al., 1964) as delineated by the valley walls. For subsequent lower terraces, the original area was delineated as the valley floor area minus the area occupied by the higher terrace. This method, therefore, necessitates two assumptions: (i) the terrace filled the entire valley trough, and (ii) valley width has remained constant over the time frames of terrace development. Both are reasonable starting points for this analysis. In addition, the approach cannot account for the possible existence of stacked or row terraces (cf Lewin and Gibbard, 2010), which are not visible topographically. Results are therefore best considered as a minimum number of terraces.
Terrace preservation was calculated across the whole of the study reach ( x) to assess the overall differences in preservation of terraces between catchments. In addition, terrace preservation was calculated for successive 1-km subreach segments
ACCEPTED MANUSCRIPT
within study reaches ( x) to assess within-valley variations in terrace preservation and relationships with catchment variables.
Analysis of terraces in the field and using LiDAR DEM revealed vertical erosion on terrace treads by concentrated overland flow leading to gullying. Terrace surface dissection was therefore described as the percentage of terrace remnants showing surface dissection (TD). In addition, mean remnant dissection density was calculated from the total length (km) of the dissected network within a remnant as a ratio of remnant area (km2).
3.2.2. Configuration
The spatial relationships between terrace remnants were assessed according to (i) cross-valley alignment assessed as paired if remnants are longitudinally aligned on both sides of the valley or unpaired if longitudinally staggered, (ii) downvalley continuity, which was expressed as the valley length over which terrace remnants are preserved, normalised to the total study reach length (km km-1), and (iii) lateral connectivity, expressed as the percentage of terrace remnants that are ‘connected’ to or ‘disconnected’ from the valley sides.
3.3. Catchment variables and relationships to TPI
Surrogate variables representing drivers of terrace erosion and resistance (force-resistance framework) were derived and regressed against x to explore relationships with preservation extent or lack thereof. The surrogate erosion variables include contributing catchment area (CA), unit catchment area (UCA), drainage density (DD), total stream power (), and specific stream power (ω) that all relate to the generation of discharge and its conveyance. Derivation methods for variables are
ACCEPTED MANUSCRIPT
outlined in Table 2 with further details of stream power estimation provided below. The surrogate resistance variables are the proportional area of metamorphic (LM),
basalt (LB), and granite (LG) lithologies, valley width (VW), and percent silt/clay of the
alluvial material (MA) and underlying colluvial material (MC) (Table 2). Information on alluvium percent silt + clay is based on preexisting soils data (Murtha, 1986, 1989; Cannon et al., 1992; Murtha et al., 1996). Percent silt + clay of colluvium was inferred from basal material in sediment cores. Variables were derived for each 1-km segment along each study reach (Table 2).
Specific stream power (ω) provides a measure of the rate of energy expenditure per unit area of channel (W m-2) and is defined as
ω = Qs / w (3)
where is unit weight of water (9800 N m-3), Q is discharge (m3 s-1), s is the channel slope (m m-1), and w is the mean channel width (m).
Estimates of Q were based on 1.5-year average recurrence interval (ARI) flood (Q1.5), which is considered representative of bankfull discharge in this region (Wallace et al., 2012). The 1.5-year ARI flood was derived from flood frequency analysis of catchment stream flow records using a common record length across all catchments (which in this study was 37 years; source: http://watermonitoring.derm.qld.gov.au/host.htm). As the downstream location of stream flow gauges varied between catchments, study reach Q1.5 was adjusted using the scaling factor F, where the ratio of flows at the gauged and ungauged locations was assumed equal to the ratio of upstream contributing areas to the power of b (Grayson et al., 1996) using
ACCEPTED MANUSCRIPT
where AC is the ungauged contributing area (calculated for each segment as above),
AG is the gauged contributing area, and b is 0.7 (Grayson et al., 1996). Discharge for
each catchment study area (Q1.5C) was then calculated as
Q1.5C = F*Q1.5G (5)
Channel slope (s) was calculated from the SRTM over a distance of 1 km (500 m upstream and downstream) as used by Jain et al. (2006) in their stream power calculations for the Hunter River in southeast Australia. Channel width (w) was calculated as the average width of the currently active channel within each 1-km segment (Table 2). Space for time substitution was used to infer that contemporary stream powers are representative of those over the longer term (Paine, 1985). While there are inherent assumptions with this approach, it is considered acceptable for broad indications of landscape change (Fryirs et al., 2012).
Nonparametric Spearman correlation coefficients (rs values) were computed between TPITi and all surrogate variables, given the majority of variables showed
nonnormal distributions. Significance was defined at p < 0.01.
4. Results
4.1. Terrace identification and regional correlation 4.1.1. Spatial extent
Based on the initial desktop mapping, ~5.5 km2 of terraces were mapped in the Liverpool catchment, ~6.5 km2 in the North and South Johnstone catchments, ~8 km2 in the Daintree catchment, and ~18.5 km2 in the Tully catchment. Terraces are fragmented in the Daintree, North and South Johnstone, and Liverpool catchments while being more continuous in the Tully catchment.
ACCEPTED MANUSCRIPT
4.1.2. Longitudinal profilesLongitudinal profiles for terraces, floodplains, and the present channel are illustrated in Fig. 2 and distinguish the different terrace surfaces in each catchment. In the Daintree, Liverpool, and South Johnstone catchments, terraces are on average 10-15 m above the channel. Two terraces are each in the North Johnstone and Tully catchments that differ in height by 2 to 4 m. In the North Johnstone, the higher terrace comprises a single remnant (Fig. 2B), which although insufficient for generating a longitudinal profile, was clearly distinguished in the field from the lower terrace surface. With the exception of the North Johnstone where the long profiles are more convex and differ somewhat between the terrace, floodplain, and channel, longitudinal profiles in the other catchments are similar. Long profiles are steeper in catchments of the North Johnstone, South Johnstone, Liverpool, and Tully compared to those in the Daintree.
ACCEPTED MANUSCRIPT
Fig. 2. Longitudinal profiles of terraces, floodplains, and channels in study reaches in the (A) Daintree, (B) North Johnstone, (C) South Johnstone, (D) Liverpool, and (E) Tully catchments. Average gradients (S) of terraces are shown and soil types (Virgil, Tully, Innisfail) occurring on terrace remnant surfaces symbolised. Note the varying scale of elevation (m AHD) on the y-axis.
ACCEPTED MANUSCRIPT
4.1.3. Soil typeThe three distinct soil types that occur across the terraces are the Virgil, Tully, and Innisfail soils (Murtha, 1986, 1989; Cannon et al., 1992; Murtha et al., 1996). The ‘Virgil’ soil is a red (5YR), massively structured, sandy clay loam with a distinct earthy fabric and is classified as a Red Kandosol. The Tully soil is a yellow-brown (10YR) structured silty clay loam and is a classified as a Yellow Dermosol. The Innisfail soil is a brown (7.5YR) clay loam and is a classified as a Brown Dermosol. All three soils are described as having relatively uniform texture profiles with no major particle size change with depth over the 2 m profiles.
The distribution of these mapped soils is plotted relative to the terrace elevation in Fig. 2 and illustrates that each terrace is typically associated with a single soil type. However, the higher terraces in the Daintree and Tully catchments comprise two mapped soil types (Virgil and Tully soils). In these two catchments, the Virgil soil type is mapped upstream of the Tully soil type (Fig. 2). The single remnant of the higher terrace in the North Johnstone catchment is mapped as the Virgil soil type.
4.1.4. Sedimentology
Sedimentary analysis of drill cores and available exposures highlighted important similarities between characteristics of each soil type and their underlying sediments. Terrace sediments associated with the Virgil soil type were extracted from the North Johnstone catchment (Table 3). Both sediment core samples contained clay loams, which display the characteristic red-hued colours (5YR) of the Virgil soil type throughout the 7-9.5 m profiles and are clearly separated from the underlying basal clay-rich colluvium (Fig. 3).
ACCEPTED MANUSCRIPT
Terrace sediments associated with the Tully soil type were extracted from the Liverpool catchment and were drilled to refusal at 8 m, induced by compacted colluvium (Table 3). Basal units of sand and gravel deposits are overlain by fine-grained, yellow-hued (10YR) sandy clay loam sediments with an interbedded fine sand unit (Fig. 3). Mottling of fine-grained sediments to 4.5 m provides evidence for extensive weathering.
Sediments underlying the Innisfail soil type were sampled in the North Johnstone, South Johnstone, and Tully catchments. In the South Johnstone and Tully catchments, cores were drilled to depths of 5-6 m, with mottled clays interpreted here to be colluvium occurring as shallow as 2-3 m (Table 3, Fig. 3). In the North Johnstone catchment, the core was drilled to 6 m where a cobble deposit, interpreted to be channel material, prevented deeper drilling (Table 3, Fig. 3). In all of these five cores, the alluvial sediments consisted of fine-grained, brown (7.5YR) sandy clay loams, which are similar to the properties of the Innisfail soil. The consistent brown colour of all sediments underlying the Innisfail soil across these three catchments suggests similar weathering regimes.
ACCEPTED MANUSCRIPT
Fig. 3. Representative core logs and OSL ages (ka) associated with different mapped soil terrace types of (A) Virgil soil, (B) Tully soil, and (C) Innisfail soil.
4.1.5. Terrace chronology
A preliminary terrace chronology is provided by OSL dating of near-surface and basal material within terraces, as well as from basal material in floodplains across four catchments (Table 4). No terrace or floodplain material was obtained from the
ACCEPTED MANUSCRIPT
Daintree catchment. Sampled terraces were associated with the three different soil types and are used to constrain the age of abandonment and period of terrace removal. Of the six terrace samples submitted from the Virgil soil in the North Johnstone catchment, statistically significant ages could not be determined for five of the samples owing to a lack of dateable quartz. The only age produced from this profile is derived from a sample at 1.3 m, which was dated to 14.4 ± 1.6 ka (Table 4). Samples from sediments associated with the Innisfail soil in the North Johnstone catchment produced an age of 7.75 ± 0.71 ka (5 m), while the sample at 1.3 m was dated to 1.22 ± 0.14 ka. Basal floodplain units were dated to 0.94 ± 0.10 ka. Based on these dates, the timing of abandonment and subsequent removal for terraces comprising the Virgil soil can be constrained to between 14 and 7.75 ka and between 1.22 and 0.94 ka for terraces of the Innisfail soil.
Sediments underlying the Tully soil in the Liverpool catchment yielded ages of 13.7 ± 2.3 ka (2.2 m), 12.5 ± 2.1 ka (4.2 m), and 12.9 ± 2.2 ka (2.5 m) (Table 4). Basal floodplain units could not be dated because of insufficient dateable quartz. The OSL dates from terraces are all within error of each other and indicate that the earliest age of abandonment is ~13.3 ka (based on the mean age of two near-surface samples). This age is similar to the abandonment age of ~14.4 ka for sediments associated with the Virgil soil in the North Johnstone.
Sediments underlying the Innisfail soil were sampled in the South Johnstone, North Johnstone, and Tully catchments. Samples from South Johnstone yielded basal ages of 7.06 ± 0.60 ka (4.4 m), 7.15 ± 0.89 ka (2.5 m), 7.75 ± 1.00 ka (3.5 m) (Table 4). Similar basal ages were derived in the North Johnstone dated at 7.59 ± 1.09 ka (4.5 m). Of the three near-surface samples from the South Johnstone, only one was able to be dated and produced an age of 5.29 ± 0.49 ka (1.3 m); it is similar to the age
ACCEPTED MANUSCRIPT
of 4.48 ± 0.48 ka (1.3 m) derived from near-surface material in the Tully catchment. In combination with basal floodplain ages of 1.11 ± 0.15 ka in the South Johnstone, 1.56 ± 0.13 ka in the Tully, and 0.94 ± 0.10 ka in the North Johnstone, the earliest age of abandonment of terraces that comprise the Innisfail soil can be constrained to between 5.29 and 1.11 ka.
4.1.6. Regional terrace correlations
Elevation and soil type classifications of terraces, verified using sedimentary data and the OSL chronology, were used to correlate terraces across the five study catchments. Consequently, two terraces were recognised regionally. The higher T1 terrace encompasses the Virgil and Tully soils, which although pedologically distinct, were dated to have similar ages of abandonment occurring at 13.3 ka in the Liverpool catchments and 14.4 ka in the North Johnstone catchment. These cannot be distinguished based on the standard deviations of OSL ages (Table 4). The period of removal of the T1 terrace occurred until 7.4 ka when the T2 terrace began infilling. The timing of the infilling of T1 is poorly constrained by this chronology. The T1 terrace occurs in the Daintree, North Johnstone, Liverpool, and Tully catchments (Figs. 4 and 5).
The lower T2 terrace is characterised by the Innisfail soil, which is found to be chronologically distinct (Table 4). Infilling occurred at ~7.4 ka (average of basal ages from North and South Johnstone catchments), while terrace abandonment occurred at ~4.9 ka (mean of near-surface terrace ages in South Johnstone and Tully catchments), with removal occurring until ~1.2 ka (mean of basal floodplain ages in North Johnstone, South Johnstone, and Tully catchments). The relatively young age derived for the near-surface sediments in the North Johnstone catchment (1.2 ka) most likely
ACCEPTED MANUSCRIPT
indicates deposition during overbank flooding around this time. The T2 terrace occurs in the North and South Johnstone and the Tully catchments (Figs. 4 and 5).
Fig. 4. Maps showing the distribution of regionally correlated terraces and representative valley cross sections in study catchments of (A) Daintree River and (B)
ACCEPTED MANUSCRIPT
Tully River. Terrace remnant classifications by soil type are shown: Vi = Virgil; Tu = Tully; In = Innisfail.
Fig. 5. Maps showing the distribution of regionally correlated terraces and representative valley cross sections in study catchments of (A) North Johnstone River, (B) South Johnstone River, and (C) Liverpool Creek. Terrace remnant classifications
ACCEPTED MANUSCRIPT
4.2. Terrace preservationTerrace preservation as described by the terrace preservation index (TPIx), terrace surface dissection, and remnant configurations are summarised for each terrace and catchment in Table 5. Considerable variability exists in T1 terrace preservation between catchments, which range from 0.38 in the Daintree and Liverpool catchments to 0.07 in the North Johnstone catchment. In contrast, preservation of the T2 terrace is relatively similar between catchments (TPIT2 = 0.44-0.61). Comparisons of TPIT1 and TPIT2 across catchments shows that preservation is inversely proportional with highest T1 preservation where the T2 terrace is absent and conversely, highest T2 preservation where T1 extents are minimal. The majority of T1 and T2 terrace remnants are dissected, although gully networks are most developed in the T1 terraces in the Daintree and Liverpool catchments (Fig. 6).
Intercatchment variability in terrace configuration is considerable. For example, T1 terrace remnants in the Daintree catchment are predominantly unpaired while paired remnants predominantly characterise the T2 terrace in the Tully catchment. The majority of terrace remnants are ‘connected’ to the valley side, or higher terraces in the case of the T2 terrace in the Tully catchment. However, mapping also revealed the existence of ‘disconnected’ terrace remnants in the Liverpool catchment (Fig. 5C). Disconnected terrace remnants are isolated from the valley sides and appear in profile as ‘mesas’ and are surrounded by floodplains that display evidence of a network of abandoned channels.
ACCEPTED MANUSCRIPT
Fig. 6. LiDAR DEM images showing representative examples of gullies that dissect T1 terrace surfaces in the Daintree (A) and Liverpool catchments (B). Terrace fronts are denoted by the solid white line and valley margin by the dashed white line.
4.3. Relationship between terrace preservation and catchment variables
Correlation analysis (Spearman, rs) showed that, surrogate variables for drivers of terrace erosion and resistance are not strongly related to the preservation of the T1 terraces ( T1) or the T2 terraces ( T2)(Table 6). Specific stream power (ω) was the only signifcant variable (p < 0.01) correlated with T1 and with T2 and shows a negative correlation. Visualisation of bivariate scatterplots revealed some clear clusters in the data that corresponded to each catchment. Thus, to examine these relationships further T1 and T2 were plotted against the variables of ω, MA, LB,
and LM, which we consider to represent the key drivers of erosion and resistance.
Figure 7 shows the median and 10th and 90th percentiles for T1 and T2 plotted against the four selected erosion/resistance variables. Similar patterns are evident between T1 and T2 and variables of ω and LB, although these are characterised
ACCEPTED MANUSCRIPT
by positive and negative trends respectively. Associations with MA and LM differ
ACCEPTED MANUSCRIPT
Fig. 7. Plots showing catchment median values and 10th and 90th percentiles of T1 (left column) and T2 (right column) and (A) specific stream power (ω), (B) terrace alluvium percent silt + clay (MA), (C) proportion basalt lithology (LB), and (D)
proportion metamorphic lithology (LM). DT – Daintree, LI – Liverpool, NJ – North
Johnstone, SJ – South Johnstone, TL – Tully.
5. Discussion
Mapped, correlated, and dated terraces form the basis for our understanding of how river systems evolve (Leopold et al., 1964). River terraces are known to be preserved in a wide range of settings and basin sizes (Macklin et al., 2002; Bridgland and Westaway, 2008) and have been described here across five catchments in the Wet Tropics region of northeast Australia. This study identified, and classified, dominant terraces across the region, with initial emphasis on available topographic and soils data. The use of surface elevations in terrace classifications is well established (Leopold et al., 1964) and availability of digital elevation models to derive elevation information is now common (Stokes et al., 2012). Similarly, the properties of alluvial soils are considered to reflect time-dependant processes and are commonly used in the identification (Birkeland, 1990; Huggett, 1998) and relative dating of terraces (Tsai et al., 2007). Combining long profiles and soil characteristics was implicit in the early notion of ‘K’ weathering cycles (Walker, 1962) and used to develop age-to-height relationships, which were then interpreted as synchronous regional climatic phases (Warner, 1970, 1972; Young, 1976).
Elevation and soils data sets were useful starting points for the identification of terraces across the region; but when verified with independent sedimentary and chronological data, a number of limitations were identified. This suggests no single
ACCEPTED MANUSCRIPT
attribute should be relied upon for regional correlation of terraces. For example, terraces with a common elevation of ~10 m were mapped in four of the five catchments, but the additional data (e.g., soils, dating) identified these as different terraces (e.g., T1 in the Tully and T2 in South Johnstone). Chronological data proved critical for the final terrace correlations. Although several samples did not return a statistically significant age because of low populations of quartz, reported ages were sufficient to confirm the distinct timing of abandonment of the T1 (7.4-13.9 ka) and T2 terraces (1.2-4.9 ka). The OSL data also indicated that different soil types (i.e., Virgil and Tully), although displaying clear pedological differences, are of similar age (13-14 ka). These two soil types co-occur in the Daintree and Tully catchments with the sandier Virgil soils existing upstream of the fine-grained Tully soils. Their different characteristics might represent a downstream fining as a result of overall downstream reductions in stream power (Jain et al., 2008).
The age-to-height and chronostratigraphic data presented appears relatively straightforward compared with previous investigations in similar partly confined valleys in New South Wales (NSW) (Cohen and Nanson, 2008; Cheetham et al., 2010). These studies highlighted the existence of ‘polycyclic’ terraces and cautioned that alluvial units in these narrow valley settings can share similar elevations but be very different in basal age. Thus, the assumption that the continuity of terrace surfaces along a valley represents coeval formation is often invalid. A recognised limitation of this study, like much previous work, is the spatial representation of the chronostratigraphic sections used to verify the elevation and soils data. Many sites are restricted to the lower end of the long profile. However, terrace soils were found to be distinct across the topographic surfaces. Soils are regional in their extent, show similar weathering characteristics to deeper sediments, and represent two
ACCEPTED MANUSCRIPT
chronologically distinct groups thereby providing some assurance that the chronology is representative of terraces throughout the region. Where valleys retain ‘patchy’ terrace remnants, the resultant age-height data sets will be complex compared to those resulting from more whole-of-valley removal of alluvial units. The notable absence of T1 or T2 terraces from any given catchment seems a likely indicator of complete removal from these valleys. Further dating is required to investigate these interpretations. For example, a recent study in the Daintree catchment reported the existence of three terrace surfaces (Leonard and Nott, 2015), with the highest terrace, correlating well in terms of elevation (10-15 m above the channel) and age (18-30 ka) with the T1 terrace identified in this study (Thomas et al., 2007; Leonard and Nott, 2015). The additional middle terrace comprises a single, small remnant ~50 m wide, and the lower terrace (basal ages of ~1 ka) correlates well with what is mapped in this study as the annually inundated floodplain surface. This existing study confirms, however, the broad trends reported for the other Wet Tropics catchments in the current study and confirms that existing soils mapping accurately characterised the underlying sediments to 7-m depth (Leonard and Nott, 2015).
5.1. Terrace preservation
Terrace preservation reflects the nature and rate of river processes since terrace formation (Lewin and Macklin, 2003). The terrace sediments dated in this study indicate an alluvial record spanning the 4-15 ka time interval, which is within the lower range of the alluvial record from similar partly confined valleys in NSW (10-35 ka) (Cohen and Nanson, 2008; Cheetham et al., 2010). However, the record detailed here is considerably shorter and younger than those from the larger and wider valleys west of the Great Dividing Range (Page et al., 1996), coastal southeastern
ACCEPTED MANUSCRIPT
catchments (Nott et al., 2002; Nanson et al., 2003), and northern dry tropics of the Fitzroy basin (Croke et al., 2011). As noted elsewhere, the relatively steep, narrow, and partly confined nature of the settings in the current study enhances the removal of alluvium through higher stream power. High stream power, as a result of the regular occurrence of high magnitude-low frequency extreme flood events and cyclones (Nott and Hayne, 2001), is likely to explain the poor preservation potential of alluvial material in these tropical catchments. Older terraces in this study are generally less preserved (i.e., lower TPI) than younger terraces. This is in contrast to other investigations that note the inherent bias in the preservation of older terraces thought to be more resistant to erosion (Lewin and Macklin, 2003; Lewin et al., 2005; Cheetham et al., 2010). Preservation of T1 terraces has an important influence on the preservation potential of subsequent alluvium. This was demonstrated in this study by the absence of a T2 terrace where T1 is well preserved (Daintree and Liverpool catchments) and vice versa, with higher extents of the T2 terrace where T1 is spatially restricted (e.g., Tully catchment).
Quantitative explanations for the variability in terrace preservation using an array of catchment variables designed to reflect variations in removal and resistance processes proved inconclusive at the regional scale. Correlation analyses highlighted the potentially important roles of specific stream power and underlying lithology, but these are relatively weak indicators of terrace preservation across the region (rs < 0.6) (Table 6). While to some degree this was influenced by catchment-specific patterns, there were several other confounding factors. Firstly, the resolution of the data sets used to derive some of the variables such as stream power was, by necessity, coarse given available data and large spatial extent of five study catchments. Notably, the use of contemporary discharge records as representative of discharges over timeframes of
ACCEPTED MANUSCRIPT
terrace development could be questioned. However grain size records in core logs indicated a relatively small variability in the size of transported material across the valley floors. Unlike terraces in other settings in NSW where coarse gravel lags over bedrock were taken to indicate much higher discharge regimes between late Pleistocene terrace and Holocene floodplain units (Nanson and Young, 1987, 1988; Cohen and Nanson, 2008), the dominance of silts and clays in these sequences overlapping thick regolith mantles would suggest that relatively low stream powers were associated with flooding and alluvial deposition in the valley trough. The age of abandonment of the T1 terrace post-dates the Last Glacial Maximum (LGM) (21–18 ka) that was characterised as relatively dry and cool in northeastern Australia with estimated reductions in rainfall of 60% reported (Kershaw and Nanson, 1993; Moss and Kershaw, 2000), which indicates the infilling of this terrace may have occurred during the LGM. The second possible cause of poor correlations is the likely influence of extreme discharge events in the removal of alluvial sequences. Dated terraces confirm large gaps in alluvial sedimentation that is best explained by the episodic removal of central tracts of valley floor in response to high magnitude flood events. These processes are poorly represented in the range of variables used in the analysis.
5.2. Terrace preservation and force-resistance conditions
While surrogate variables for erosion and resistance were not strongly correlated with terrace preservation across the region (Table 6), differences between catchments were evident (Fig. 7). This highlights the potential of a range of different force-resistance processes driving terrace preservation. To examine this further we focused on the three major types of terrace configurations: paired, unpaired, and
ACCEPTED MANUSCRIPT
disconnected (Table 5). We evaluated how these may be explained using the force-resistance variables of stream power (ω) and alluvial material force-resistance (MA). For each configuration type, values for the explanatory variables were derived from the catchment where the configuration is best exemplified. From the analysis of bivariate plots (Fig. 7), lithology variables LB and LM were excluded based on their absence
from some catchments, which limited their use as discriminatory variables at the regional scale. The resulting relationship between the three terrace configuration types, force-resistance variables and terrace preservation ( x) is shown in Fig. 8. This provides some insight into how force-resistance variables interact to influence terrace preservation.
Fig. 8. Major types of terrace configurations plotted according to specific stream power and percent silt + clay of terrace alluvium. Points are categorised by median
x.
5.3. Terrace typology and a conceptual model of terrace preservation
Characterisation of the dominant terrace configurations and their relationship to drivers of erosion and resistance within the fluvial system (Fig. 8) allows some
ACCEPTED MANUSCRIPT
commentary on the processes driving the preservation of the three major terrace types (Fig. 9).
Fig. 9. Terrace types and dominant processes of preservation.
5.3.1. Type 1: Paired, overlapped
Cross-valley pairing of T1 and T2 terrace remnants is evident in all study catchments, particularly in the lower reaches; however, this configuration is best characterised by the T2 terrace in the Tully catchment where paired remnants occupy the majority of the valley in upstream and downstream reaches. Sediments drilled from paired terraces across four catchments revealed the presence of colluvium in the base of cores at all sites. A lack of erosional contacts and grain size variability with depth indicate that terrace alluvium is onlapped over older colluvial valley fills. Specifically, in the Tully catchment, soils developed on fine-grained colluvium flank the entire study reach (Murtha, 1986). The depth to bedrock underlying the colluvium could not be established owing to the resistant nature of the colluvial sediments, but previous studies in the region report common depths of at least 4-6 m (Thomas et al., 2007). The material properties and characteristics of the overlying fluvial deposits indicate that colluvial regolith likely forms an important source of fine-grained material for the construction of these overlapped terraces in this tropical setting.
ACCEPTED MANUSCRIPT
Paired terraces are used as an indicator for the relative importance of vertical channel incision over lateral channel adjustments (Bull, 1979; Merritts et al., 1994), which enhance the preservation potential of terraces (Lewin and Macklin, 2003). Paired terrace configurations dominate in settings with relatively high specific stream powers and alluvium that is of relatively moderate resistance (Fig. 8). While the relatively cohesive alluvium would provide some lateral resistance to erosion during bankfull and overbank events, it is the presence of colluvium underlying the terraces that is also important for the preservation of this terrace type. The lower resistance of colluvial deposits to downcutting, compared to that of highly resistant granite bedrock (Pazzaglia et al., 1998), enhances the effectiveness of vertical incisional processes. Once incised, the highly cohesive banks constrain lateral fluvial adjustment and undercutting of terrace alluvium. The influence of colluvium on terrace preservation highlights the importance of inherited catchment morphologies on subsequent fluvial development in these catchments (Phillips, 2006; Fryirs and Brierley, 2010). Vertical incision may occur gradually in association with annual flooding events in these rivers. However, more rapid episodes of vertical incision may occur in association with extreme discharges given the sensitivity of channel long profiles in tectonically inactive, regolith-dominated settings (Pazzaglia et al., 1998). In this region, extreme discharges are most likely related to the occurrence of floods associated with monsoonal depressions and cyclones.
5.3.2. Type 2: Unpaired, inset
Unpaired terraces are most common in the upper reaches of the study catchments, particularly the Daintree catchment, where remnants typically are small. Remnants are larger in the lower reaches of the North Johnstone catchment. This form
ACCEPTED MANUSCRIPT
is underlain by channel deposits as was observed in the North Johnstone, and this suggests these forms are associated with laterally mobile channels (Merritts et al., 1994; Lewin and Gibbard, 2010). In this study, unpaired terraces are associated with lower stream powers and lower resistance of alluvial material (Fig. 8). However, stream powers in these study reaches are considerably higher than is typical of laterally migrating systems (cf Nanson and Croke, 1992). This together with the morphology of the adjacent floodplains — which show flood chutes, abandoned channels, and backswamps — suggests that removal of these terrace forms occurs primarily through rapid channel adjustments. These terraces types are typically connected to the valley side margins and are often characterised by highly dissected surfaces (Fig. 6) highlighting the additional role of hillslope runoff processes in the development of gully networks that enhance terrace removal.
5.3.2. Type 3: Disconnected
Disconnected terrace remnants are isolated mesa-type forms most commonly
observed in the Liverpool Creek catchment. The form indicates that removal processes operate at both the terrace front and valley sides through the repeated occurrence of channel avulsion as reported for similar settings elsewhere (Erskine et al., 2005, 2012; Phillips, 2012). Imagery from the Liverpool floodplain clearly indicates the presence of multiple palaeochannels and flood chutes, and it is likely such channels operate effectively at removing terrace material from the valley margins and terrace front. Extensive dissection of the terrace surfaces in the Liverpool catchment, as indicated by well-developed gully networks, many of which are concentrated at the terrace-hillslope margin, suggests that this is an additional process that contributes to the removal of terrace remnants from these valleys.
ACCEPTED MANUSCRIPT
6. Conclusion
Understanding how rivers have adjusted to variations in past climate and associated discharge and sediment supply is critical for understanding and predicting how these systems will respond to future changes in climate and rainfall. Alluvial terraces are important landforms that provide information on how catchments have adjusted over time. This study described the characteristics and spatial preservation of terraces in five catchments in tropical northeast Queensland, Australia. Long profiles, soil type, sedimentology, and OSL dating were used to identify and correlate two terraces (T1, T2) across the study catchments. The preservation of T1 and T2 terraces were described using the terrace preservation index (TPI), and remnant configuration used to infer dominant removal processes. At the regional scale, quantitative explanations for terrace preservation, based on an array of surrogate variables for erosion and resistance, revealed relatively weak correlations. However, catchment-specific differences in terrace preservation were evident. Combining this information with major terrace configurations (paired, unpaired, disconnected) and TPI identified that a combination of erosion and resistance factors are responsible for driving terrace preservation. This study provides important insights into terrace preservation in a humid tropical environment, filling a significant research gap in a region that is likely to undergo significant change in future climate and land use (Wohl et al., 2012). This study provides a framework for future research linking the timing of fluvial response to palaeoclimate change in the Wet Tropics.
ACCEPTED MANUSCRIPT
Table 1Characteristics of selected catchments and study reaches
Catchment characteristics Study reach characteristics
River Area (km2) Max elevation (m) Mean annual rainfall range (mm) Dominant lithology Main channel length (km) Valley length (km) Mean valley width (km) Average channel slope (m m-1) Mean channel width (m) Sinuosity Mean floodplain width (km) Daintree 1327 1345 1500-4000 Metamorphic 129 19 1.05 0.0009 50 1.38 0.61 North Johnstone 1030 1365 2400-4000 Basalt 112 19 0.60 0.0020 70 1.17 0.38 South Johnstone 545 1160 2400-4800 Basalt, Granite 92 17 0.78 0.0020 60 1.45 0.40 Liverpool 320 1145 2800-4800 Metamorphic 48 15 0.97 0.0021 25 1.34 0.64 Tully 1684 1205 2400-4800 Granite 130 20 1.43 0.0019 50 1.20 0.57
ACCEPTED MANUSCRIPT
Table 2Selected catchment variables and data sources
Table 3
Terrace sedimentary characteristics
Catchment Soil type (regional terrace) Core ID Drilling depth (m) Sedimentary characteristics North Johnstone Virgil (T1)
N1 9.5 Dark reddish brown (5YR3/3) sandy clay loam.
Mottles (red, grey) increasing with depth. No bedding or distinct boundaries observed within this unit. Clear transition to clay, with veins of sand and gravel (colluvium) at 9.3 m.
N2 7.0 Yellowish red (5YR4/6) sandy clay loam.
Mottles from 4 m (red, grey). No bedding or distinct boundaries observed within this unit.
Variable Description Data source
Contributing
catchment area, CA (km2)
Sum area of upstream SRTM-derived subcatchments
SRTM DEM (30 m)
Unit catchment area, UCA
CA / Total catchment area (km2) SRTM DEM (30 m) Drainage density,
DD (km km-2)
Total channel length (km) within CA / CA SRTM DEM (30 m); 1: 250, 000 drainage mapping (Source: Queensland Government). Specific stream power, ω (W m-2) ω = Qs / w
(see text for derivation method)
As above for total stream power, average channel width measured from satellite imagery (n = 10 per 1-km segment).
Lithology LM LB LG
Metamorphic area within CA / CA Basalt area within CA / CA Granite area within CA / CA
1: 250, 000 Geology mapping (Source: Geoscience Australia), SRTM DEM (30 m)
Valley width, Vw Mean valley width* (km) / CA
*(n = 10)
Alluvial valley floor as defined by 1:50, 000 soils mapping (Source:
Queensland Government) Terrace alluvial
material percent silt + clay, MA (%)
% silt + % clay 1:50, 000 soils mapping
(Source: Queensland Government) Colluvial material
percent silt + clay, MC (%)
% silt + % clay Sediment core basal
material sampled in this study
ACCEPTED MANUSCRIPT
(T2) distinct boundaries observed within this unit.
Gradual transition to sand from 4.5 m. Clear boundary to cobbles at 5.5 m.
South Johnstone
Innisfail (T2)
S1 5.0 Brown (7.5YR4/4) sandy clay loam. No
bedding or distinct boundaries observed within this unit. Clear boundary to mottled (grey, orange, red) clay (colluvium) at 4.5 m.
S2 5.0 Brown (7.5YR4/3) sandy clay loam. No
bedding or distinct boundaries observed within this unit. Clear boundary to mottled (grey, orange, red) clay (colluvium) at 3.0 m.
S3 10.0 Brown (7.5YR4/4) sandy clay loam. No
bedding or distinct boundaries observed within this unit. Clear boundary to mottled (grey, orange, red) clay (colluvium) at 4.5 m. Liverpool Tully
(T1)
L1 8.0 Yellowish brown (10YR5/6), mottled sandy
clay loam with thin (0.2 m) fine sand unit. Clear boundary to medium sand at 4.5 m. Clear boundary to gravel (<4 mm) at 5.2 m. Clear boundary to coarse sand at 5.7 m. Clear
boundary to mottled clay (colluvium) at 7.0 m.
Tully Innisfail
(T2)
Tu1 6.0 Brown (7.5YR4/3) clay loam. No bedding or
distinct boundaries observed within this unit. Mottled clays with angular fine (2 mm) gravels and coarse sand (colluvium) from 2.0 m.
Table 4
Sample details and OSL ages (2 σ); ‘np’ indicates where an OSL age could not be returned
Catchment Soil type
(regional terrace) Core ID Sample depth (m) Dose rate (Gy ka-1)
De (Gy) Age (ka)
North Johnstone Virgil (T1) N1 1.3 3.09 ± 0.28 44.4 ± 2.8 14.4 ± 1.6 6.0 2.32 ± 0.20 n/a np 9.0 2.59 ± 0.23 n/a np N2 1.3 3.08 ± 0.27 n/a np 2.5 3.85 ± 0.37 n/a np 6.5 2.31 ± 0.24 n/a np Innisfail (T2) N3 1.3 4.03 ± 0.34 4.93 ± 0.40 1.22 ± 0.14 5.0 3.36 ± 0.27 26.0 ± 1.1 7.75 ± 0.71 Floodplain N4 6.5 3.53 ± 0.30 3.32 ± 0.23 0.94 ± 0.10 South Johnstone Innisfail (T2) S1 1.3 5.20 ± 0.45 27.5 ± 0.8 5.29 ± 0.49 4.4 4.14 ± 0.33 29.2 ± 0.9 7.06 ± 0.60
ACCEPTED MANUSCRIPT
S2 1.0 4.17 ± 0.44 n/a np 2.5 3.35 ± 0.35 23.9 ± 1.6 7.15 ± 0.89 S3 1.8 4.37 ± 0.36 n/a np 3.5 4.25 ± 0.38 32.9 ± 3.0 7.75 ± 1.00 4.5 4.38 ± 0.43 33.3 ± 3.5 7.59 ± 1.09 Floodplain S4 6.0 3.33 ± 0.27 3.69 ± 0.40 1.11 ± 0.15 Liverpool Tully (T1) L1 2.2 4.66 ± 0.39 64.0 ± 9.5 13.7 ± 2.3 L1 4.2 3.67 ± 0.27 45.9 ± 7.0 12.5 ± 2.1 L2 2.5 3.96 ± 0.33 51.2 ± 7.6 12.9 ± 2.2 L2 7.5 3.10 ± 0.27 n/a np Floodplain L3 6.3 3.11 ± 0.21 n/a np Tully Innisfail (T2) Tu1 1.3 4.95 ± 0.43 22.2 ± 1.8 4.48 ± 0.48 Floodplain Tu2 5.5 3.26 ± 0.25 5.07 ± 0.19 1.56 ± 0.13ACCEPTED MANUSCRIPT
Table 5Summary table of terrace topographic and preservation characteristics in study reaches
Terrace Average elevation above channel (m) Mean slope (m m-1) TPIx Surface dissection (gully density, km km-2) Configuration Cross-valley alignment Lateral connectivity Downvalley connectivity (km km-1) Daintree T1 14 0.0008 0.38 90% dissected (5.0) 18% paired 100% connected 0.70
North Johnstone T1 12 Na – single
remnant only 0.07 100% dissected - single remnant only (1.1) Unpaired – single remnant only 100% connected 0.05 T2 10 0.0017 0.44 50% dissected (2.6) 55% paired 100% connected 0.80
South Johnstone T2 10 0.0017 0.45 30% dissected
(2.6) 45% paired 100% connected 0.50 Liverpool T1 10 0.0019 0.35 50% dissected (5.2) 58% paired 60% connected 0.50 Tully T1 10 0.0018 0.10 100% dissected (2.6) 50% paired 100% connected 0.25 T2 6 0.0011 0.61 100% dissected (1.0) 88% paired 100% connected 0.85
ACCEPTED MANUSCRIPT
Table 6Spearman correlation coefficients computed between terrace preservation for the T1 terrace ( T1) (n = 74) and T2 terrace ( T2) (n = 58) preservation and surrogate catchment variables for drivers of terrace erosion (CA, UCA, DD, ω) and resistance (LM, LG, LB, VW, MA, MC) (refer to Table 2); values in bold are significant at p < 0.01
CA UCA DD ω LM LG LB VW MA MC
T1 0.26 0.39 0.35 -0.47 0.47 0.07 -0.51 0.18 0.09 -0.14
ACCEPTED MANUSCRIPT
AcknowledgementsThis work was supported by the Wet Tropics Management Authority Student Research Scheme (grant no. 862). The authors thank Peter Todd (Queensland Department of Natural Resources and Mines) for assistance with LiDAR processing; Queensland Government DSITI staff for assistance with fieldwork and sample processing, particularly Jeremy Manders and Ian Hall for their drilling expertise; and Jon Nott for sharing his understanding of the region. Thanks also to landholders for permission to work on their properties, sharing their knowledge, and helpful assistance during fieldwork. We are also grateful for the editorial assistance of Richard Marston and comments from three anonymous reviewers who greatly improved this manuscript.
References
Aitken, M.J., 1998. An Introduction to Optical Dating: The Dating of Quaternary Sediments by the Use of Photon-Stimulated Luminescence. Oxford University Press, Oxford, 267 pp.
Birkeland, P.W., 1990. Soil-geomorphic research – a selective overview. Geomorphology 3(3-4), 207-224.
Bonell, M., 1988. Hydrological Process of the Wet Tropical Coast of NE Queensland. In: Warner, R.F. (Ed.), Fluvial geomorphology of Australia. Academic Press, Sydney, pp. 41-68.
Bonell, M., Gilrnour, D.A., 1980. Variations in short-term rainfall intensity in relation to synoptic climatological aspect of the humid tropical northeast Queensland coast. Singapore J. Trop. Geo. 1(2), 16-30.
ACCEPTED MANUSCRIPT
Bowler, J.M., Johnston, H., Olley, J.M., Prescott, J.R., Roberts, R.G., Shawcross, W., Spooner, N.A., 2003. New ages for human occupation and climatic change at Lake Mungo, Australia. Nature 421(6925), 837-840.
Bridgland, D., Westaway, R., 2008. Climatically controlled river terrace staircases: A worldwide Quaternary phenomenon. Geomorphology 98(3-4), 285-315. Brierley, G., 1996. Channel morphology and element assemblages: a constructivist
approach to facies modelling. In: Carling, P., Dawson, M.R. (Eds.), Advances in fluvial dynamics and stratigraphy. Wiley Interscience, Chichester, pp. 263-298.
Bull, W.B., 1979. Threshold of critical power in streams. Geol. Soc. Am. Bull. 90(5), 453.
Bull, W.B., 1990. Stream-terrace genesis: implications for soil development. Geomorphology 3(3-4), 351-367.
Cannon, M.G., Smith, C.D., Murtha, G.G., 1992. Soils of the Cardwell-Tully Area, North Queensland. Report no. 115. CSIRO Publishing, Australia, 147 pp. Cheetham, M.D., Bush, R.T., Keene, A.F., Erskine, W.D., 2010. Nonsynchronous,
episodic incision: Evidence of threshold exceedance and complex response as controls of terrace formation. Geomorphology 123(3-4), 320-329.
Cohen, T.J., Nanson, G.C., 2008. Topographically associated but chronologically disjunct late Quaternary floodplains and terraces in a partly confined valley, south-eastern Australia. Earth Surf. Process. Landf. 33(3), 424-443.
Colman, S.M., 1983. Progressive changes in the morphology of fluvial terraces and scarps along the Rappahannock River, Virginia. Earth Surf. Process. Landf. 8(3), 201-212.
